Japanese prosecutors use generative AI in virtual asset investigation

일본 검찰이 가상자산 거래 추적 수사 등 일부 업무에 생성형 인공지능(AI)을 활용할 수 있도록 했다.

According to NHK on the 13th, Japan's Supreme Prosecutors' Office prepared these guidelines last month and delivered them to frontline prosecutors' offices in line with the Japanese government's policy to expand the use of generative AI by public institutions such as central ministries.

The guidelines include two major tasks where generative AI can be used.

The task is to summarize or translate the relevant content when papers in specialized fields such as investigation and medicine that track virtual asset transaction flows are needed for investigations and trials.

Recently, in Japan, there have been a series of cases in which branch-type criminal groups recruited through social networking services (SNS) are laundering criminal profits using virtual assets to avoid investigation.

Prosecutors expect that the introduction of generative AI will allow them to quickly identify increasingly complex financial flows.

However, the guidelines also specify strict precautions when using generative AI, considering the risk of information leakage or incorrect answers.

First, the AI ​​that can be used was limited to a few pre-specified models, and the accuracy and objectivity of the generated results had to be verified.

In addition, it was requested not to enter highly confidential investigative information and to use generative AI only as a reference and auxiliary tool for work.